RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

Abacus AI

Product
The name of the base system (platform): Artificial intelligence (AI, Artificial intelligence, AI)
Developers: Abacus
Date of the premiere of the system: August 2021
Branches: Information technology

2021: Launch a real-time enterprise machine learning platform

In mid-August 2021, Abacus launched a platform that, according to the company itself, is the world's first solution for machine learning and real-time deep learning on an enterprise scale.

The Abacus platform allows large and small companies to broadcast real-time events such as clicks, online purchases, social networking interactions and media views from websites and the Internet of Things metric. This data can then be processed, transformed, and used to train deep learning models capable of generating real-time context predictions.

The world's first enterprise real-time machine learning platform Abacus AI is launched

The Abacus platform includes simple configuration of data pipelines, tools for cleaning and converting data, training and hosting models, monitoring and explaining the model, as well as a real-time storage of machine learning functions. Users can train their own models using the Abacus neural network architecture, or they can use popular PyTorch or TensorFlow frameworks along with the novelty.

Abacus AI users can begin to get acquainted with the platform with a simple use case, such as a recommendation mechanism, gradually expanding the capabilities. There is an option to use a low-code platform that developers can use to solve simpler artificial intelligence problems.

For more experienced operators and enterprises, Abacus simplifies the creation of more complex models by using application programming interfaces - writing Python code and SQL queries. Data experts can use their own models on the platform.

It is also possible to use the Abacus AI engine to create individual deep learning models for specific use cases and data sets. In this case, the Abacus platform will choose the most suitable neural architecture from a number of types of neural networks, including LSTM, RNN, variational autocoders.[1]

Notes